Add rake task to check integrity of uploaded files

1 files changed, 44 insertions, 0 deletions

diff --git a/lib/tasks/gitlab/uploads.rake b/lib/tasks/gitlab/uploads.rake
new file mode 100644
index 00000000000..fd5ef78103b
--- /dev/null
+++ b/lib/tasks/gitlab/uploads.rake
@@ -0,0 +1,44 @@
+namespace :gitlab do
+  namespace :uploads do
+    desc 'GitLab | Uploads | Check integrity of uploaded files'
+    task check: :environment do
+      puts 'Starting checking integrity of uploaded files'
+
+      uploads_batches do |batch|
+        batch.each do |upload|
+          puts "- Checking file (#{upload.id}): #{upload.absolute_path}".color(:green)
+
+          if upload.exist?
+            check_checksum(upload)
+          else
+            puts "  * File does not exist on the file system".color(:red)
+          end
+        end
+      end
+
+      puts 'Done!'
+    end
+
+    def batch_size
+      ENV.fetch('BATCH', 200).to_i
+    end
+
+    def calculate_checksum(absolute_path)
+      Digest::SHA256.file(absolute_path).hexdigest
+    end
+
+    def check_checksum(upload)
+      checksum = calculate_checksum(upload.absolute_path)
+
+      if checksum != upload.checksum
+        puts "  * File checksum (#{checksum}) does not match the one in the database (#{upload.checksum})".color(:red)
+      end
+    end
+
+    def uploads_batches(&block)
+      Upload.all.in_batches(of: batch_size, start: ENV['ID_FROM'], finish: ENV['ID_TO']) do |relation| # rubocop: disable Cop/InBatches
+        yield relation
+      end
+    end
+  end
+end